MIMII-Agent: Leveraging LLMs with Function Calling for Relative Evaluation of Anomalous Sound Detection

2025-07-28

AI Summary

To address the challenge of conducting reliable cross-machine-type relative performance evaluation of unsupervised anomaly sound detection (UASD) systems in the absence of genuine anomalous audio, this paper proposes a large language model (LLM)-driven text-to-anomaly-sound synthesis method. Given textual fault descriptions, the method leverages LLM function-calling capabilities to automatically select and compose audio transformations, generating semantically coherent and acoustically plausible machine-type-specific anomalous sounds from normal audio. Crucially, it requires no real anomalous samples or model fine-tuning, thereby significantly improving scalability and realism of UASD evaluation. Experiments demonstrate strong consistency (Spearman ρ > 0.85) between the detection difficulty rankings of synthesized versus real anomalies across multiple UASD models, validating the synthesized anomalies as effective proxies for relative performance assessment.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search

Abstract

This paper proposes a method for generating machine-type-specific anomalies to evaluate the relative performance of unsupervised anomalous sound detection (UASD) systems across different machine types, even in the absence of real anomaly sound data. Conventional keyword-based data augmentation methods often produce unrealistic sounds due to their reliance on manually defined labels, limiting scalability as machine types and anomaly patterns diversify. Advanced audio generative models, such as MIMII-Gen, show promise but typically depend on anomalous training data, making them less effective when diverse anomalous examples are unavailable. To address these limitations, we propose a novel synthesis approach leveraging large language models (LLMs) to interpret textual descriptions of faults and automatically select audio transformation functions, converting normal machine sounds into diverse and plausible anomalous sounds. We validate this approach by evaluating a UASD system trained only on normal sounds from five machine types, using both real and synthetic anomaly data. Experimental results reveal consistent trends in relative detection difficulty across machine types between synthetic and real anomalies. This finding supports our hypothesis and highlights the effectiveness of the proposed LLM-based synthesis approach for relative evaluation of UASD systems.

Problem

• Generating machine-specific anomalies without real anomaly data
• Overcoming unrealistic sounds from manual label-based augmentation methods
• Using LLMs to create diverse anomalies from normal sounds

Innovation

• LLMs interpret fault descriptions for anomalies
• Automated audio transformation for anomaly synthesis
• Relative evaluation without real anomaly data